Dynamic Inference of Likely Symbolic Tensor Shapes in Python Machine Learning Programs
In machine learning programs, it is often tedious to annotate the dimensions of shapes of various tensors that get created during execution. We present a dynamic likely tensor shape inference analysis, called ShapeIt, that annotates the dimensions of shapes of tensor expressions with symbolic dimension values and establishes the symbolic relationships among those dimensions. Such annotations can be used to understand the machine learning code written in popular frameworks, such as PyTorch and JAX, and to find bugs related to tensor shape mismatch. We have implemented ShapeIt on top of a novel dynamic analysis framework for Python, called Pynsy, which works by instrumenting Python bytecode on the fly. Our evaluation of ShapeIt on several tensor programs illustrates that ShapeIt could effectively infer symbolic shapes and their relationships for various neural network programs with low runtime overhead.
Thu 18 AprDisplayed time zone: Lisbon change
11:00 - 12:30 | Testing 3Research Track / Journal-first Papers / Software Engineering in Practice at Grande Auditório Chair(s): José Miguel Rojas The University of Sheffield | ||
11:00 15mTalk | Do Automatic Test Generation Tools Generate Flaky Tests? Research Track Martin Gruber BMW Group, University of Passau, Muhammad Firhard Roslan University of Sheffield, Owain Parry The University of Sheffield, Fabian Scharnböck University of Passau, Phil McMinn University of Sheffield, Gordon Fraser University of Passau Pre-print | ||
11:15 15mTalk | Deep Combination of CDCL(T) and Local Search for Satisfiability Modulo Non-Linear Integer Arithmetic Theory Research Track Xindi Zhang Institute of Software Chinese Academy of Science, Bohan Li Institute of Software Chinese Academy of Science, Shaowei Cai Institute of Software at Chinese Academy of Sciences | ||
11:30 15mTalk | Uncover the Premeditated Attacks: Detecting Exploitable Reentrancy Vulnerabilities by Identifying Attacker Contracts Research Track Shuo Yang Sun Yat-sen University, Jiachi Chen Sun Yat-sen University, Mingyuan Huang Sun Yat-Sen University, Zibin Zheng Sun Yat-sen University, Yuan Huang School of Data and Computer Science, Sun Yat-sen University, Guangzhou, China | ||
11:45 15mTalk | Practical Non-Intrusive GUI Exploration Testing with Visual-based Robotic Arms Research Track Shengcheng Yu Nanjing University, Chunrong Fang Nanjing University, Mingzhe Du Nanjing University, Yuchen Ling Nanjing University, Zhenyu Chen Nanjing University, Zhendong Su ETH Zurich | ||
12:00 15mTalk | Dynamic Inference of Likely Symbolic Tensor Shapes in Python Machine Learning Programs Software Engineering in Practice Pre-print | ||
12:15 7mTalk | Mutation Analysis for Evaluating Code Translation Journal-first Papers Giovani Guizzo Brick Abode, Jie M. Zhang King's College London, Federica Sarro University College London, Mark Harman Meta Platforms, Inc. and UCL, Christoph Treude Singapore Management University | ||
12:22 7mTalk | Generalized Coverage Criteria for Combinatorial Sequence Testing Journal-first Papers Achiya Elyasaf Ben-Gurion University of the Negev, Eitan Farchi IBM Haifa Research Lab, Oded Margalit Ben-Gurion University of the Negev, Gera Weiss Ben-Gurion University of the Negev, Yeshayahu Weiss Ben-Gurion University of the Negev Link to publication DOI | ||